sustainability

Article

Research on Factors Influencing MunicipalHousehold Solid Waste Separate Collection: BayesianBelief NetworksZhujie Chu 1,2,*, Wenna Wang 1,2, Bairong Wang 3,* and Jun Zhuang 3

1 School of Economics and Management, Harbin Engineering University, Harbin 150001, China;hegangwwn@163.com

2 Research Institute of Disaster and Crisis Management, School of Economics and Management,Harbin Engineering University, Harbin 150001, China

3 Department of Industrial and Systems Engineering, University at Buffalo, State University of New York,317 Bell Hall, Buffalo, NY 14260-2050, USA; jzhuang@buffalo.edu

* Correspondence: chuzhujie@hrbeu.edu.cn (Z.C.); bairongw@buffalo.edu (B.W.);Tel.: +86-451-8251-9916 (Z.C.); +1-716-645-4707 (B.W.)

Academic Editor: Marc A. RosenReceived: 23 December 2015; Accepted: 2 February 2016; Published: 5 February 2016

Abstract: Municipal household solid waste (MHSW) has become a serious problem in China overthe course of the last two decades, resulting in significant side effects to the environment. Therefore,effective management of MHSW has attracted wide attention from both researchers and practitioners.Separate collection, the first and crucial step to solve the MHSW problem, however, has not beenthoroughly studied to date. An empirical survey has been conducted among 387 households inHarbin, China in this study. We use Bayesian Belief Networks model to determine the influencingfactors on separate collection. Four types of factors are identified, including political, economic,social cultural and technological based on the PEST (political, economic, social and technological)analytical method. In addition, we further analyze the influential power of different factors, basedon the network structure and probability changes obtained by Netica software. Results indicatethat technological dimension has the greatest impact on MHSW separate collection, followed by thepolitical dimension and economic dimension; social cultural dimension impacts MHSW the least.

Keywords: municipal household solid waste; separate collection; influence factors; Bayesian beliefnetworks model

1. Introduction

Municipal Household Solid Waste (MHSW) in China has increased significantly due to the rapideconomic development. It not only causes serious pollution problems, which are not conducive toenvironment and human health [1,2], but also retards sustainable development of society [3]. It willprolong the processing cycles if MHSW is not been separated properly. The MHSW collected andtransportation in China was 172 million tons in 2013. In addition, that number reached 179 milliontons in 2014. However, there were still 15 million tons of MHSW that had not been harmlesslytreated in 2014, which caused a series of environmental problems [4,5]. This is mainly because ofunsatisfactory levels of MHSW separate collection, which causes poor effectiveness of the subsequentprocessing. MHSW separate collection is the prerequisite for the MHSW reduction [6] and causesfewer environmental problems than other solutions [7]. Hence it is of high importance in effectivelymitigating the environmental problems caused by MHSW [8,9]. However, many problems makeMHSW separate collection hard to be popularized [10]. Examples include the complicated compositionof MHSW, its large production, difficulties in treatment, low environmental awareness of the public,

Sustainability 2016, 8, 152; doi:10.3390/su8020152 www.mdpi.com/journal/sustainability

Sustainability 2016, 8, 152 2 of 14

an imperfect management system, etc. In order to improve the effectiveness of MHSW separatecollection, it is necessary to analyze the influencing factors on separate collection, which will in turnreturn significant guidance in achieving the optimal resource utilization of MHSW [11].

2. Literature Review

Existing research has shown that economic, political, social conditions and the attitudes of peopleare necessary for recycling success [12–19]. Guagnano et al. [20] concluded that all the external factorsincluding psychology, finance, law and society may bring a positive or negative impact on the residents’behavior. Noehammer and Byer [21] founded that compulsory recycling programs launched by thegovernment had a higher participation rate than voluntary resident recycling. Rada et al. [22] proposedthat the MHSW legislation at the national and international level was necessary for the planning ofgood MHSW management. DETR [23] stated that all actions of the UK Government were aimed topromote the implementation of MHSW reduction, and also took steps to reduce the amount of MHSW.The Strategy Unit [24] also stressed the importance of governmental related actions to stem the spreadof MHSW.

A number of studies have shed light on the significance of financial incentives in MHSW separatecollection. Everett and Peirce [25] suggested that the financial incentives were the direct impetus toMHSW separate collection behaviors. Ragazzi et al. [26] pointed out that incentives might give peoplea strong sense of collaboration, which could motivate more people to get involved in the fight againstMHSW. Glenn [27] and Steuteville [28] showed that the success of separate collection depended onmarket conditions. They also suggested that government should make relevant incentive policies toencourage public participation. Noh et al. [29] also found that penalties could stimulate participatinginitiative of people who did not do separate collection. Meanwhile rewarding people who perform wellin separate collection also motivates the public greatly. Yau [30] suggested that economic incentivesworked well in promoting MHSW separate collection in Hong Kong.

Environmental attitudes of the public matters greatly in MHSW separate collection. By usingempirical analysis, Minton and Rose [31] concluded that the environmental attitudes were thedeterminant of MHSW separate collection behaviors. A man who cared about nature was morelikely to engage in separating collection. Desa et al. [32] pointed out that the knowledge level, theattitude and the environmental consciousness of the public impacted MHSW separate collection.Miafodzyeva and Brandt [33], Afroz et al. [34] considered that the participation and willingnessof the public were the key influential factors in MHSW separate collection. However, Zhang andWen [35] hold an opposite opinion that attitudes and willingness did not have significant impacts onMHSW separate collection behavior in the case study at Suzhou, China. As for the public propagandaand education, Wellar and Barry [36] and Guerra [37] performed a thorough study. They showedthat propaganda could motivate residents to realize the significance of MHSW separate collectionand hence perform MHSW separate collection better. De Feo and De Gisi [38] presented the ideathat public education and citizen encouragement could reduce MHSW in the design of separatecollection processes.

The MHSW collection also serves matter, like the quantity, species and locations of equipment.Enough supply of assorted dustbins could improve MHSW cyclic utilization. Derksen and Gartrell [39]studied the residents' behaviors of MHSW separate collection, and found that people would classifyMHSW if the dustbins were convenient to reach in the communities. Even people who werenot concerned about the environment would perform recycling too, because of the convenience.Hence, the convenience of MHSW separate collection is also one of the decisive factors to promoteseparate collection of MHSW, as shown by Timlett and Williams [40], Ando and Gosselin [41] andSidique et al. [42]. Bernstad [43] found that besides the convenience, infrastructure construction wasquite necessary to separate collection.

In conclusion, existing research shows that influencing factors on MHSW separate collection canvary from policy, law, finance, culture, etc., which laid the foundation for this study. However, there

Sustainability 2016, 8, 152 3 of 14

are no comprehensive studies regarding MHSW separate collection in China, the biggest developingcountry, with the largest population in the world. Harbin is in the northeast region of China, itsunique features of climate makes the MHSW separate collection more pertinent. The research resultson MHSW separate collection in the region are universal, which can be conducive to achieve theminimization, recycling and harmlessness of MHSW. Bayesian Belief Networks (BBNs) model hasbeen used as an effective tool for uncertainty events analysis [44,45] and gains a great advantage ininfluencing factors investigation research, such as the one in this study. Therefore, together with thePEST analytical method and system theory, we analyze the influential factors of MHSW separatecollection from four dimensions: political, economic, social cultural and technological dimension.

3. Data and Methodology

3.1. Study Area

Harbin, the capital city of the Heilongjiang province, is an important political, economic andcultural center in northeastern China with its abundant natural resources such as forests, minerals, etc.It also has the largest land area and the second largest population in municipalities of China, whichmakes our investigation for Harbin more representative and general. Studied districts are illustratedin Figure 1 [46].

Sustainability 2016, 8, 152  3 of 15 

In  conclusion,  existing research shows that  influencing factors on MHSW  separate  collection 

can vary from policy, law, finance, culture, etc., which laid the foundation for this study. However, 

there  are  no  comprehensive  studies  regarding MHSW  separate  collection  in  China,  the  biggest 

developing country, with the largest population in the world. Harbin is in the northeast region of 

China,  its  unique  features  of  climate makes  the MHSW  separate  collection more  pertinent.  The 

research results on MHSW separate collection in the region are universal, which can be conducive to 

achieve the minimization, recycling and harmlessness of MHSW. Bayesian Belief Networks (BBNs) 

model has been used as an effective  tool  for uncertainty events analysis  [44,45]  and gains a great 

advantage  in  influencing  factors  investigation  research,  such as  the one  in  this  study. Therefore, 

together with the PEST analytical method and system theory, we analyze the  influential factors of 

MHSW  separate  collection  from  four  dimensions:  political,  economic,  social  cultural  and 

technological dimension. 

3. Data and Methodology 

3.1. Study Area 

Harbin,  the capital city of  the Heilongjiang province,  is an  important political, economic and 

cultural center in northeastern China with its abundant natural resources such as forests, minerals, 

etc. It also has  the  largest  land area and  the second  largest population  in municipalities of China, 

which makes our  investigation  for Harbin more  representative and general.  Studied districts are 

illustrated in Figure 1 [46]. 

 

Figure 1. Studied districts of Harbin city. 

Harbin has a population of 9,952,100 in 2013, with an urban population of 4,736,300. The city of 

Harbin collects and transports about 1,314,000 tons of MHSW each year [47]. Generation quantities 

and  the  composition  of MHSW  change with  seasons  and months.  For  example, more MHSW  is 

produced by inhabitants in the winter than summer due to the generation of heating waste, such as 

coal ash. At  the same  time, MHSW  is more  in January, February, and October  than other months 

because of holiday activities. MHSW generated by households  is  temporarily  stored  in bins, and 

then sanitation vehicles provide two collection services every day to collect and transport MHSW in 

the area. However, there is obviously a distinction between the numbers of MHSW bins in different 

communities. Senior communities have more MHSW bins. Each building has one bin which ensures 

the public make separate collections  in a convenient and efficient way. Nevertheless, for ordinary 

communities,  several  buildings  share  one  collection. MHSW  are  collected  by  sanitation workers 

door‐to‐door. 

3.2. Data Collection 

Both  simple  random  sampling  and  stratified  sampling  were  used  during  the  selection  of 

investigated household families. 

Figure 1. Studied districts of Harbin city.

Harbin has a population of 9,952,100 in 2013, with an urban population of 4,736,300. The city ofHarbin collects and transports about 1,314,000 tons of MHSW each year [47]. Generation quantitiesand the composition of MHSW change with seasons and months. For example, more MHSW isproduced by inhabitants in the winter than summer due to the generation of heating waste, such ascoal ash. At the same time, MHSW is more in January, February, and October than other monthsbecause of holiday activities. MHSW generated by households is temporarily stored in bins, and thensanitation vehicles provide two collection services every day to collect and transport MHSW in the area.However, there is obviously a distinction between the numbers of MHSW bins in different communities.Senior communities have more MHSW bins. Each building has one bin which ensures the public makeseparate collections in a convenient and efficient way. Nevertheless, for ordinary communities, severalbuildings share one collection. MHSW are collected by sanitation workers door-to-door.

3.2. Data Collection

Both simple random sampling and stratified sampling were used during the selection ofinvestigated household families.

Families are selected from the districts of Nan-gang, Dao-li, Dao-wai, Xiang-fang and Ping-fang.There were five communities selected in each district and then sixteen household families were

Sustainability 2016, 8, 152 4 of 14

surveyed with questionnaires in each community. In this survey, family members who were familiarwith MHSW separate collection finished the questionnaires.

Five groups of students from Harbin Engineering University participated in the research surveyduring 12–26 March 2015. Data was obtained mainly by door-to-door visits. Families with retiredmembers were visited in the morning and those with working members were surveyed in the afternoonor lunchtime. Phone survey techniques were adopted for families who missed the first survey visits.387 out of the 400 sampled households were surveyed in this study, yielding a 96.75% completion rate(the questionnaire is available in Appendix A).

3.3. Influential Factors of MHSW Separate Collection

3.3.1. Political Dimension

China is a government-leading country which means all levels of governments are responsiblefor issuing legislation on MHSW separate collection. Dimensions for politics is denoted by P, and itsparent nodes are: (1) The enforcement of laws; (2) waste classification indicators; and (3) supervisionintensity of governments, denoted by P1, P2 and P3, respectively.

3.3.2. Economic Dimension

Financial support motivates MHSW separate collection, and hence two factors, infrastructureinvestment and support for involved companies, are identified for economic dimension, expressed asE1 and E2, respectively. Investment in infrastructure includes MHSW collection buckets and transportvehicles. Support for relevant companies refers to governmental subsidization to companies dealingwith MHSW, such support includes tax reduction, direct financially support, etc.

3.3.3. Sociocultural Dimension

Social culture has evolved in the long-term historical development of human society andexerts influence on people’s behaviors. In this study, three specific factors comprising socioculturaldimensions are identified: (1) public environmental awareness, denoted by S1; (2) intensity of publicity,denoted by S2; (3) importance community has attached into MHSW separate collection, denoted by S3.

3.3.4. Technological Dimension

Advanced technology lays the foundation for the effectiveness of MHSW separate collection andfacilitates the smooth going of its operation. We divided the technological dimension into two parts.The first one refers to technology and equipment in MHSW disposal centers and is denoted by T1.The second factor is the construction of a MHSW disposing center and is denoted by T2.

3.4. Bayesian Belief Networks for Influence Factors

3.4.1. Bayesian Belief Network Model

Bayesian Belief Networks (BBNs), based on probability analysis and graph theory, serve as botha great inference pattern and a description of uncertain knowledge. It is represented by directed acyclicgraphs illustrating dependence between variables. Specifically, nodes in the graph represent all randomvariables studied, and the directed arcs show conditional dependencies between these variables.

A Bayesian Belief Network consists mainly of the topological structure (G) and the parameter (θ).Therefore the learning of BBNs contains structure learning and parameter learning which refers toconstructing the directed acyclic structure and obtaining the related conditional probabilistic tables(CPTs), respectively.

The mathematical basis of BBNs is the Bayesian equation, as shown below in Equation (1):

P pA|Bq “P pB|Aq P pAq

P pBq(1)

Sustainability 2016, 8, 152 5 of 14

Two variables of a BBN, G and θ, confirm the only distribution among the nodes X1, ¨ ¨ ¨ , Xm.Each joint distribution can be represented by the conditional probability. To each variable Xi, wecan calculate its conditional probability under all direct or indirect control of variables X1, ¨ ¨ ¨ , Xi´ 1.The joint probability distribution for X “ tX1, ¨ ¨ ¨ , Xnu is given by the chain rule:

P pX1, ¨ ¨ ¨ , Xnq “

nź

i“1

P pXi|X1, ¨ ¨ ¨ , Xi´1q (2)

On the basis of Markov hypothesis for structures, every variable Xi depends only on the nodePαi to which its arrow is towards. The joint distribution in Equation (2) could be simplified further asfollowing in Equation (3):

P pX1, ¨ ¨ ¨ , Xnq “

nź

i“1

P`

Xiˇ

ˇPαi

˘

(3)

3.4.2. Structure Learning in BBNs

In this study, we obtain a BBN with 15 nodes and 14 lines as shown in Figure 2. Arrows connectingthe nodes show dependencies between variables and probabilities on lines represent the extent ofdependencies. Three types of nodes are identified in this network, namely root nodes, parent nodesand child nodes. The single one root node refers to the final effectiveness of MHSW separate collection.Parent nodes represent sub-factors from each influencing factor dimension. Children nodes mainlyrefer to factors from the four dimensions. For instance, the parent nodes of Economic dimension (E)are infrastructure investment (E1) and financial support (E2), the child node depends on its parentnodes and it directly affects effectiveness of MHSW.

chain rule:

1 1 11

, , | , ,n

n i ii

P X X P X X X

(2)

On the basis of Markov hypothesis for structures, every variable iX depends only on the node

iP to which its arrow is towards. The joint distribution in equation 2 could be simplified further

as following in equation 3:

11

, , |i

n

n ii

P X X P X P

(3)

3.4.2 Structure learning in BBNs

In this study, we obtain a BBN with 15 nodes and 14 lines as shown in Figure 2. Arrows

connecting the nodes show dependencies between variables and probabilities on lines represent

the extent of dependencies. Three types of nodes are indentified in this network, namely root

nodes, parent nodes and child nodes. The single one root node refers to the final effectiveness of

MHSW separate collection. Parent nodes represent sub-factors from each influencing factor

dimension. Children nodes mainly refer to factors from the four dimensions. For instance, the

parent nodes of Economic dimension ( E ) are infrastructure investment ( 1E ) and financial support

( 2E ), the child node depends on its parent nodes and it directly affects effectiveness of MHSW.

Figure2. BBN structure for factors influencing MHSW separate collection

3.4.3 Parameter learning in BBNs

Specifically, nodes’ contributions are divided into five levels: “excellent”=5, “good”=4,

“medium”=3, “bad”=2 and “extremely bad”=1. According to standards in each level, participants

will score MHSW service, separate collection knowledge, etc. For example, the question that

(NO.2 in questionnaire) “How much do you know MHSW classification standards?” is used to

measure the completeness and standardization on the MHSW separate collection evaluation index

system. “Excellent” means a satisfying formal and comprehensive classification standard exists in

this area.

Construction of MHSW

processing center

Technological

dimension

Political

dimension

Waste classification

indicators

Effectiveness of MHSW

separate collection

Socioultural

dimension

Economic

dimension

Subsidization or support

for relevant companies

Intensity of publicity

Infrastructure

investment

Supervision intensity

of governments

Public environmental

awareness

Importance community

has attached into MHSW

separate collection

Enforcement of laws

Technology and equipment

in MHSW disposal center

Figure 2. Bayesian Belief Network (BBN) structure for factors influencing municipal household solidwaste (MHSW) separate collection.

3.4.3. Parameter learning in BBNs

Specifically, nodes’ contributions are divided into five levels: “excellent” = 5, “good” = 4,“medium” = 3, “bad” = 2 and “extremely bad” = 1. According to standards in each level, participantswill score MHSW service, separate collection knowledge, etc. For example, the question that (NO. 2in questionnaire) “How much do you know MHSW classification standards?” is used to measurethe completeness and standardization on the MHSW separate collection evaluation index system.“Excellent” means a satisfying formal and comprehensive classification standard exists in this area.

Children nodes are scored and divided into five levels with the aforementioned method anda score for each child node, namely P, E, S and T, is computed as the average score of its parent nodes.Therefore, the score of each child node also ranges from 1 to 5. We assume that the score is uniformly

Sustainability 2016, 8, 152 6 of 14

distributed, and divided it into (4.2, 5], (3.4, 4.2], (2.6, 3.4], (1.8, 2.6] and [1, 1.8] five intervals in total,which correspond to “excellent”, “good”, “medium”, “bad”, and “extremely bad” respectively.

The satisfaction level of “excellent” is defined by “A”, “good” is defined by “B”, “medium” isdefined by “C”, “bad” is defined by “D”, and “extremely bad” is defined by “E”.

We can obtain a Bayesian network by using Netica software with the probability distribution ofeach node. The results are shown in Figure 3.

Sustainability 2016, 8, 152  6 of 15 

3.4.3. Parameter learning in BBNs 

Specifically,  nodes’  contributions  are  divided  into  five  levels:  “excellent”  =  5,  “good”  =  4, 

“medium” = 3, “bad” = 2 and “extremely bad” = 1. According to standards in each level, participants 

will score MHSW service, separate collection knowledge, etc. For example, the question that (NO. 2 

in questionnaire) “How much do you know MHSW classification standards?” is used to measure the 

completeness  and  standardization  on  the MHSW  separate  collection  evaluation  index  system. 

“Excellent” means a satisfying formal and comprehensive classification standard exists in this area. 

Children nodes are scored and divided into five levels with the aforementioned method and a 

score for each child node, namely  P ,  E ,  S   and  T , is computed as the average score of its parent 

nodes. Therefore, the score of each child node also ranges from 1 to 5. We assume that the score is 

uniformly distributed,  and divided  it  into  (4.2,  5],  (3.4,  4.2],  (2.6,  3.4],  (1.8,  2.6]  and  [1,  1.8]  five 

intervals in total, which correspond to “excellent”, “good”, “medium”, “bad”, and “extremely bad” 

respectively. 

The satisfaction level of “excellent” is defined by “A”, “good” is defined by “B”, “medium” is 

defined by “C”, “bad” is defined by “D”, and “extremely bad” is defined by “E”. 

We can obtain a Bayesian network by using Netica software with the probability distribution 

of each node. The results are shown in Figure 3. 

 

Figure 3. BBN for factors influencing MHSW separate collection. 

Each  node possesses a  conditional probabilistic  table  (CPT)  showing  relations between each 

node and  its parents. Part of the CPT  for node  R   (effectiveness of MHSW separate collection)  is 

illustrated in Figure 4. The left side of the table shows different combinations of its parent nodes  P , 

E ,  S   and  T .  The  right  side  shows  the  conditional  probability  distribution  of  node  R   given different combinations of its parent nodes. 

Figure 3. BBN for factors influencing MHSW separate collection.

Each node possesses a conditional probabilistic table (CPT) showing relations between each nodeand its parents. Part of the CPT for node R (effectiveness of MHSW separate collection) is illustratedin Figure 4. The left side of the table shows different combinations of its parent nodes P, E, S and T.The right side shows the conditional probability distribution of node R given different combinations ofits parent nodes.

Sustainability 2016, 8, 152  7 of 15 

 

Figure 4. Part of CPT for node  R . 

4. Results and Discussion 

In order  to  learn better  information about  the  relations between  the  effectiveness of MHSW 

separate  collection  and  its  influencing  factors, we  discuss  results  by  reverse  inference,  namely 

comparing changes in satisfying level distributions of each dimension given two different satisfying 

levels of final effectiveness of separate collection. Two cases where effectiveness of MHSW separate 

collection (R ) shows “excellent”, scored at 5 and “medium”, scored at 3, are studied. The network 

reasoning results are shown in Figures 5 and 6 respectively. 

 

Figure 5. Bayesian network parameter graph for  R   = 5. 

Figure 4. Part of CPT for node R.

Sustainability 2016, 8, 152 7 of 14

4. Results and Discussion

In order to learn better information about the relations between the effectiveness of MHSWseparate collection and its influencing factors, we discuss results by reverse inference, namelycomparing changes in satisfying level distributions of each dimension given two different satisfyinglevels of final effectiveness of separate collection. Two cases where effectiveness of MHSW separatecollection (R) shows “excellent”, scored at 5 and “medium”, scored at 3, are studied. The networkreasoning results are shown in Figures 5 and 6 respectively.

Sustainability 2016, 8, 152  7 of 15 

 

Figure 4. Part of CPT for node  R . 

4. Results and Discussion 

In order  to  learn better  information about  the  relations between  the  effectiveness of MHSW 

separate  collection  and  its  influencing  factors, we  discuss  results  by  reverse  inference,  namely 

comparing changes in satisfying level distributions of each dimension given two different satisfying 

levels of final effectiveness of separate collection. Two cases where effectiveness of MHSW separate 

collection (R ) shows “excellent”, scored at 5 and “medium”, scored at 3, are studied. The network 

reasoning results are shown in Figures 5 and 6 respectively. 

 

Figure 5. Bayesian network parameter graph for  R   = 5. Figure 5. Bayesian network parameter graph for R = 5.Sustainability 2016, 8, 152  8 of 15 

 

Figure 6. Bayesian network parameter graph for  R   = 3. 

The satisfaction  level distribution of each node given  final separate collection effectiveness  is 

excellent  ( R   =  5)  and  is  significantly  different  from  that  when  final  effectiveness  is  medium   

( R   =  3).  Distributions  for  politics  dimension  ( P ),  economy  dimension  ( E ),  social  cultural 

dimension ( S ) and technology dimension (T ) all change significantly, which are shown in Figures 

7–10  respectively.  The  “ABCDE”  represent  five  satisfaction  levels  ranging  from  excellent  to 

extremely bad (“excellent” = A, “good” = B, “medium” = C, “bad” = D, and “extremely bad” = E). 

Four  graphs  refer  to  the  satisfaction  level  distribution  of  node  P ,  1P , 

2P   and 

3P   respectively 

(shown Figure 7a–d). Three graphs express  the satisfaction  level distribution of node  E ,  1E   and 

2E   respectively  (shown  Figure  8a–c).  Four  graphs  refer  to  the  satisfaction  level distribution  of 

node  S , 1S , 

2S   and 

3S   respectively  (shown  Figure  9a–d).  Three  graphs  represent  the 

satisfaction level distribution of node  T , 1T   and 

2T   respectively (shown Figure 10a–c). 

Compared with  the  state when  R   equals  to  3, when  R   equals  to  5,  the probability when 

political dimension  (P )  is “medium” decreased  from 51.2%  to 11.1%;  the probability when  P   is “good”  performance  increased  from  32.5%  to  44.3%  and  the  probability when  P   is  “excellent” performance  increased  from 6.49%  to 44.5%. According  to  the maximum subordination principle, 

the condition  ranges  from “medium”  to “good” and “excellent” almost nearly.  It means  that  the 

political dimension has an important impact on the effectiveness of MHSW separate collection. 

Specifically,  when  satisfaction  levels  of  1P   (Enforcement  of  laws),  restrictions  to  relevant 

policies  and  regulations  are  “bad”  and  “medium”  respectively.  The  probabilities  respectively 

decrease  from 17.9% and 37.8%  to 6.46% and 26.7%. While  the probabilities when  1P   are “good” and “excellent” increase from 29.6% and 12.4% to 36.3% and 30.1% respectively. When satisfaction 

levels of  2P   (Waste classification indicators), the normative degree of evaluation index system are 

“bad” and “medium” respectively. The probabilities decrease  from 13.7% and 40.4% to 3.57% and 

23.1% respectively. The probability when  2P   are “good” and “excellent” respectively increase from 

34.1%  and  10.5%  to  47.6%  and  25.2%. When  satisfaction  levels  of  3P   (Supervision  intensity  of governments),  government  regulations  and  input  are  “good”  and  “excellent”  respectively,  the 

probabilities increase from 28.4% and 14.5% to 34.7% and 34.4% respectively. 

Figure 6. Bayesian network parameter graph for R = 3.

Sustainability 2016, 8, 152 8 of 14

The satisfaction level distribution of each node given final separate collection effectiveness isexcellent (R = 5) and is significantly different from that when final effectiveness is medium (R = 3).Distributions for politics dimension (P), economy dimension (E), social cultural dimension (S) andtechnology dimension (T) all change significantly, which are shown in Figures 7–10 respectively. The“ABCDE” represent five satisfaction levels ranging from excellent to extremely bad (“excellent” = A,“good” = B, “medium” = C, “bad” = D, and “extremely bad” = E). Four graphs refer to the satisfactionlevel distribution of node P, P1, P2 and P3 respectively (shown Figure 7a–d). Three graphs expressthe satisfaction level distribution of node E, E1 and E2 respectively (shown Figure 8a–c). Four graphsrefer to the satisfaction level distribution of node S, S1, S2 and S3 respectively (shown Figure 9a–d).Three graphs represent the satisfaction level distribution of node T, T1 and T2 respectively (shownFigure 10a–c).

Sustainability 2016, 8, 152  9 of 15 

These changes  indicate that a political dimension has a stronger influence on the effectiveness 

of MHSW  separate  collection.  This  improvement  benefits  from  great  efforts made  by  the  local 

Harbin government, such as more  complete policies  and  regulations.  It also clears  the rights and 

obligations among players involved, such as governments, enterprises and the public. At the same 

time,  in  order  to  standardize  the  incorrect  behaviors  of MHSW  separate  collection  by  law,  the 

government has to strengthen the policy guidance, which  is helpful in developing correct MHSW 

disposal habits of the general public. The goal of resource utilization was achieved successfully by 

all these measures, and the effectiveness of MHSW separate collection has improved obviously. 

 

Figure 7. Changes in Political Dimension between  R   = 5 and  R   = 3. 

 

Figure 8. Changes in Economic Dimension between  R   = 5 and  R   = 3. 

Figure 7. Changes in Political Dimension between R = 5 and R = 3.

Sustainability 2016, 8, 152  9 of 15 

These changes  indicate that a political dimension has a stronger influence on the effectiveness 

of MHSW  separate  collection.  This  improvement  benefits  from  great  efforts made  by  the  local 

Harbin government, such as more  complete policies  and  regulations.  It also clears  the rights and 

obligations among players involved, such as governments, enterprises and the public. At the same 

time,  in  order  to  standardize  the  incorrect  behaviors  of MHSW  separate  collection  by  law,  the 

government has to strengthen the policy guidance, which  is helpful in developing correct MHSW 

disposal habits of the general public. The goal of resource utilization was achieved successfully by 

all these measures, and the effectiveness of MHSW separate collection has improved obviously. 

 

Figure 7. Changes in Political Dimension between  R   = 5 and  R   = 3. 

 

Figure 8. Changes in Economic Dimension between  R   = 5 and  R   = 3. Figure 8. Changes in Economic Dimension between R = 5 and R = 3.

Sustainability 2016, 8, 152 9 of 14

Sustainability 2016, 8, 152  10 of 15 

 

Figure 9. Changes in Sociocultural Dimension between  R   = 5 and  R   = 3. 

 

Figure 10. Changes in Technological Dimension between  R   = 5 and  R   = 3. 

When comparing with the state when  R   = 3, in the state when  R   = 5, the probabilities when 

E   (economic dimension) are “bad” and “medium” decrease  from 23.4% and 32.8%  to 1.99% and 

8.52%  respectively.  The  probabilities when  E   are  “good”  and  “excellent”  respectively  increase from 31.4% and 6.11% to 45.7% and 43.7%, respectively. According to the maximum subordination 

principle,  the  condition  ranges  from  “medium”  to  “good”  and  “excellent”  almost nearly.  It  also 

respects  that  the  economic dimension has  an  important  impact on  the  effect of MHSW  separate 

collection. 

When  1E   (infrastructure  construction  investment)  are  “good”  and  “excellent”,  the 

probabilities when  1E   are  “good”  and  “excellent”  respective  increase  from  25.9%  and  8.74%  to 

Figure 9. Changes in Sociocultural Dimension between R = 5 and R = 3.

Sustainability 2016, 8, 152  10 of 15 

 

Figure 9. Changes in Sociocultural Dimension between  R   = 5 and  R   = 3. 

 

Figure 10. Changes in Technological Dimension between  R   = 5 and  R   = 3. 

When comparing with the state when  R   = 3, in the state when  R   = 5, the probabilities when 

E   (economic dimension) are “bad” and “medium” decrease  from 23.4% and 32.8%  to 1.99% and 

8.52%  respectively.  The  probabilities when  E   are  “good”  and  “excellent”  respectively  increase from 31.4% and 6.11% to 45.7% and 43.7%, respectively. According to the maximum subordination 

principle,  the  condition  ranges  from  “medium”  to  “good”  and  “excellent”  almost nearly.  It  also 

respects  that  the  economic dimension has  an  important  impact on  the  effect of MHSW  separate 

collection. 

When  1E   (infrastructure  construction  investment)  are  “good”  and  “excellent”,  the 

probabilities when  1E   are  “good”  and  “excellent”  respective  increase  from  25.9%  and  8.74%  to 

Figure 10. Changes in Technological Dimension between R = 5 and R = 3.

Compared with the state when R equals to 3, when R equals to 5, the probability when politicaldimension (P) is “medium” decreased from 51.2% to 11.1%; the probability when P is “good”performance increased from 32.5% to 44.3% and the probability when P is “excellent” performanceincreased from 6.49% to 44.5%. According to the maximum subordination principle, the conditionranges from “medium” to “good” and “excellent” almost nearly. It means that the political dimensionhas an important impact on the effectiveness of MHSW separate collection.

Specifically, when satisfaction levels of P1 (Enforcement of laws), restrictions to relevant policiesand regulations are “bad” and “medium” respectively. The probabilities respectively decrease from17.9% and 37.8% to 6.46% and 26.7%. While the probabilities when P1 are “good” and “excellent”increase from 29.6% and 12.4% to 36.3% and 30.1% respectively. When satisfaction levels of P2 (Waste

Sustainability 2016, 8, 152 10 of 14

classification indicators), the normative degree of evaluation index system are “bad” and “medium”respectively. The probabilities decrease from 13.7% and 40.4% to 3.57% and 23.1% respectively.The probability when P2 are “good” and “excellent” respectively increase from 34.1% and 10.5% to47.6% and 25.2%. When satisfaction levels of P3 (Supervision intensity of governments), governmentregulations and input are “good” and “excellent” respectively, the probabilities increase from 28.4%and 14.5% to 34.7% and 34.4% respectively.

These changes indicate that a political dimension has a stronger influence on the effectiveness ofMHSW separate collection. This improvement benefits from great efforts made by the local Harbingovernment, such as more complete policies and regulations. It also clears the rights and obligationsamong players involved, such as governments, enterprises and the public. At the same time, in orderto standardize the incorrect behaviors of MHSW separate collection by law, the government has tostrengthen the policy guidance, which is helpful in developing correct MHSW disposal habits of thegeneral public. The goal of resource utilization was achieved successfully by all these measures, andthe effectiveness of MHSW separate collection has improved obviously.

When comparing with the state when R = 3, in the state when R = 5, the probabilities when E(economic dimension) are “bad” and “medium” decrease from 23.4% and 32.8% to 1.99% and 8.52%respectively. The probabilities when E are “good” and “excellent” respectively increase from 31.4%and 6.11% to 45.7% and 43.7%, respectively. According to the maximum subordination principle, thecondition ranges from “medium” to “good” and “excellent” almost nearly. It also respects that theeconomic dimension has an important impact on the effect of MHSW separate collection.

When E1 (infrastructure construction investment) are “good” and “excellent”, the probabilitieswhen E1 are “good” and “excellent” respective increase from 25.9% and 8.74% to 39.2% and 17.6%respectively. When E2 (fund support) is “bad”, the probability when E2 is “bad” decreases from22.4% to 4.58%, those of “good” and “excellent” decreases from 23.8% and 11.0% to 40.8% and34.2%, respectively.

These changes indicate that economic factors have important influence on the effectiveness ofMHSW separate collection. The Harbin government has increased the investment in infrastructureconstruction for MHSW separate collection, such as purchasing more MHSW collection cans andtrucks. It is reported that the number of MHSW trucks has increased from 380 to 427, and morethan 3000 MHSW collection cans with the volume 240 L are bought. Meanwhile, the governmentprovides adequate support to facilitate the shaping of separate collection by making separate collectionpoints to make separating MHSW more convenient. Besides, the government also increased itsfinancial support. For example, it encourages the charitable donations to support development of theMHSW separate collection industry. All these measures promote the effectiveness of MHSW separatecollection obviously.

Compared with the state when R = 3, in the state when R = 5, the probabilities when S(sociocultural dimension) are “bad” and “medium” decrease from 30.2% and 43.5% to 1.06% and 16.7%respectively. The probabilities of “good” and “excellent” increase from 20.9% and 2.02% to 53.1%and 29.1% respectively. According to the maximum subordination principle, the state ranges from“medium” to “good”. It shows that the sociocultural dimension has a slight effect on the effect ofMHSW separate collection.

When S1 (public environmental awareness) is “excellent”, the probability when S1 is “excellent”increases from 14.2% to 34.0%. When S2 (strength of publicity) is “medium”, the probability when S2

is “medium” decreases from 40.0% to 26.4%. The probability “excellent” increases from 14.5% to 35.3%.When S3 (attention to community management) is “excellent”, the probability when S3 is “excellent”increases from 10.6% to 21.7%.

These changes indicate that sociocultural dimension has a slight influence on the effectiveness ofMHSW separate collection. In order to improve the effectiveness of MHSW separate collection, somecommunities in Harbin have made great efforts in increasing propaganda and education. For instance,they make full use of the television, radio, newspapers, as well as the Internet and other media forpublic education. They also motivate the public members to make every possible effort to reduce

Sustainability 2016, 8, 152 11 of 14

MHSW generation. Members of the public are encouraged to use less disposable goods to avoid waste.They not only have strong environmental awareness, but also know well about the basic knowledgeand the importance of the MHSW classification. Furthermore, some communities have adopted someincentives measures. For example, residents will be praised or rewarded for their active participationand outstanding performance in MHSW separate collection.

Compared with the state when R = 3, in the state when R = 5, the probabilities when T(technological dimension) are “bad” and “medium” decrease from 33.2% and 27.5% to 0.49% and9.00% respectively. The probabilities of “good” and “excellent” increase from 25.8% and 5.74% to 38.5%and 52.0% respectively. According to the maximum subordination principle, the state ranges sharplyfrom “bad” to “excellent”. It shows that technological dimension impacts significantly on the effect ofMHSW separate collection.

Specifically, when T1 (technology and equipment in MHSW disposal center) is “excellent”,the probability when T1 is “excellent” increases from 6.93% to 30.6%. The probability when T1

is “bad” decreases from 22.2% to 5.97%. When T2 (construction of MHSW processing center) are“good” and “excellent” respectively, the probabilities increase from 26.1% and 4.16% to 40.8% and12.1% respectively.

These changes indicate that technological dimension has a significant influence on the effectivenessof MHSW separate collection. Every factor has a positive influence on the technological dimension,and therefore impacts the effectiveness of MHSW separate collection accordingly. Up to now landfillsare most widely used in MHSW treatment in Harbin due to their low cost, but they also cause pollution.Thus, the Xiang-yang MHSW treatment plant and Xi’nanbu MHSW treatment plant have innovated theexisting technology and equipment, especially in strengthening the anti-seepage measure. Moreover,in order to manage the toxic or harmful substances which are produced by the incomplete MHSWincineration, the Heilongjiang No. 3 Electric Power Construction Corporation incinerator has carriedout the technology innovation. On the one hand, it has purchased advanced equipment. On the otherhand, it takes efforts to develop new technologies. MHSW has maximized waste recycle by applyingadvanced harmless treatment technology, and as a result pollution from MHSW treatment is alsoreduced. The fact that MHSW freezes easily during Harbin’s winter time increases the difficulties inMHSW cleaning, and the slippery conditions causes travel inconvenience to the residents. Thereforetimely MHSW collected is quite necessary. The Harbin government has increased the MHSW disposalsite settings to ensure the timely clean-up and collection of MHSW. All these efforts facilitate greaterprogress in MHSW separate collection.

These changes can be explained by the fact that MHSW in Harbin is mostly disposed in landfills.Source separation by residents is rarely necessary. Therefore technological dimension factors, liketreatment method, equipment in MHSW disposal, have the greatest impact on MHSW separatecollection. Economic dimension plays a role in MHSW separate collection in an indirect way, suchas increasing the fund support, promoting technology progress, etc. Furthermore, effectiveness ofgovernment regulations and compulsory measures are also limited in MHSW separate collection.It could motivate residents’ participation. However, improving the separation awareness of the publicis challenging. Moreover, there are lots people in Harbin earning extra income by selling recyclableMHSW. These behaviors decreased the recycling and separation collection of MHSW. Thus, thesociocultural dimension has a smaller impact on MHSW separate collection.

5. Conclusions

Many previous studies have pointed the factors that impact MHSW separate collectioneffectiveness. However, they have not provided research on developing countries, especially China.Motivated by this research gap, this paper employs the BBN model to better analyze the influencingfactors of MHSW separate collection. The four dimensions include: political, economic, socioculturaland technological. In our study, we find that all dimensions have great influence on effectivenessof MHSW separate collection by experimental survey in Harbin. The technological dimension has

Sustainability 2016, 8, 152 12 of 14

the greatest effect, followed by the political dimension and economic dimension; the socioculturaldimension has the least effect.

The model we proposed can explicitly integrate both prior knowledge and information obtainedfrom the data. BBN is well suited to our study due to its advantage to handle uncertainties. It canpresent causal relationships with estimated possibilities and consequences of nested determinants.Nevertheless, for the complexity of the model, the choices to evaluate criteria can also impact itsaccuracy. Furthermore, there is no control to people who answered the questionnaire.

However, this study is limited by unsolved questions. For example, are there any dynamicinfluencing factors on MHSW separate collection, such as climate, geology, etc.? And how do theyinfluence the MHSW separate collection? Further research can be directed toward Bayesian BeliefNetworks based on the availability of research on the MHSW separate collection. The BBN can still bedeveloped further into a dynamic model that takes into account temporal variables.

Acknowledgments: We would like to thank the editors and the anonymous reviewers for their valuable commentsand suggestions on our paper. This paper stems from a research project supported by National Nature ScienceFoundation of China (NSFC) project (No.: 71473056), National Soft Science Research Plans (No.: 2014GXS4B055),Nature Science Foundation of Heilongjiang Province (No.: G2015005), and Ministry of Education HumanitiesSocial Sciences Research Project (No.: 11YJC840046).

Author Contributions: Zhujie Chu and Wenna Wang conceived and designed the research; Wenna Wang andBairong Wang carried out the analyses, including data collection and processing; Jun Zhuang supervised theresearch and contributed to the framework of the Bayesian Belief Networks. All authors contributed to preparingand approving the final manuscript.

Conflicts of Interest: The authors declare no conflict of interest.

Appendix

Questionnaire

We sincerely hope that it will be convenient for you to fill out this questionnaire. Thank you!

Please mark “‘

” on the dimension which you think correct.

PoliticalDimension

1. Are you satisfied with the implementationof policies and regulations at present?

5-stronglysatisfied

4-satisfied 3-middle 2-unsatisfied1-stronglyunsatisfied

2. How much do you know MHSWclassification standards?

5-most4-verymuch

3-onlysome

2-a little 1-nothing

3. Are you satisfied with the governmentregulation during MHSW separate collection?

5-stronglysatisfied

4-satisfied 3-middle 2-unsatisfied1-stronglyunsatisfied

EconomicDimension

4. Are you satisfied with the investment of theinfrastructure construction including MHSWcollection cans and trucks?

5-stronglysatisfied

4-satisfied 3-middle 2-unsatisfied1-stronglyunsatisfied

5. Do you think government is necessary togive fund support to the related enterprises?

5-stronglynecessary

4-necessary 3-middle 2-unnecessary1-strongly

unnecessary

SociocultualDimension

6. Do you think it is important to make MHSWseparate collection?

5-stronglyimportant

4-important 3-middle 2-unimportant1-strongly

unimportant

7. Are you satisfied with the publicity ofMHSW separate collection through television,newspapers and Internet?

5-stronglysatisfied

4-satisfied 3-middle 2-unsatisfied1-stronglyunsatisfied

8. Are you satisfied with the attention toMHSW separate collection in yourcommunity?

5-stronglysatisfied

4-satisfied 3-middle 2-unsatisfied1-stronglyunsatisfied

TechnologicalDimension

9. Are you satisfied with the existingtechnology and equipment in MHSWdisposal?

5-stronglysatisfied

4-satisfied 3-middle 2-unsatisfied1-stronglyunsatisfied

10. Are you satisfied with the settings ofMHSW disposal sites nearby?

5-stronglysatisfied

4-satisfied 3-middle 2-unsatisfied1-stronglyunsatisfied

11. Are you satisfied with the effectiveness of MHSWseparate collection?

5-stronglysatisfied

4-satisfied 3-middle 2-unsatisfied1-stronglyunsatisfied

Sustainability 2016, 8, 152 13 of 14

References

1. Alam, R.; Chowdhury, M.A.I.; Hasan, G.M.J.; Karanjit, B.; Shrestha, L.R. Generation, storage, collectionand transportation of municipal solid waste—A case study in the city of Kathmandu, capital of Nepal.Waste Manag. 2008, 28, 1088–1097. [CrossRef]

2. Rathi, S. alternative approaches for better municipal solid waste management in Mumbai, India. Waste Manag.2006, 26, 1192–1200. [CrossRef] [PubMed]

3. Baud, I.; Grafakos, S.; Hordijk, M.; Post, J. Quality of life and alliances in solid waste management:Contributions to urban sustainable development. Cities 2001, 18, 3–12. [CrossRef]

4. National Bureau of Statistics of China (NBSC). China Statistical Yearbook; China Statistics Press: Beijing,China, 2014.

5. National Bureau of Statistics of China (NBSC). China Statistical Yearbook; China Statistics Press: Beijing,China, 2015.

6. Schiavon, M.; Ragazzi, M.; Rada, E.C.; Merler, G. Proposal for the correct management of the life cycleassessment results from integrated municipal solid waste treatment. WIT Trans. Ecol. Environ. 2014, 180,163–173.

7. Rada, E.C. Effects of MSW selective collection on waste-to-energy strategies. WIT Trans. Ecol. Environ.2013, 176, 215–223.

8. Tanaka, M. Waste management for a sustainable society. J. Mater. Cycles Waste Manag. 2007, 9, 2–6. [CrossRef]9. Kollikkathara, N.; Feng, H.; Stern, E.A. A purview of waste management evolution: Special emphasis on

USA. Waste Manag. 2009, 29, 974–985. [CrossRef] [PubMed]10. Zhang, D.Q.; Tan, S.K.; Gersberg, R.M. Municipal solid waste management in China: Status, problems and

challenges. J. Environ. Manag. 2010, 91, 1623–1633. [CrossRef] [PubMed]11. Nguyen, T.T.P.; Zhu, D.J.; Le, N.P. Factors influencing waste separation intention of residential households in

a developing country: evidence from Hanoi, Vietnam. Habitat Int. 2015, 48, 169–176. [CrossRef]12. Ball, R.; Lawson, S.M. Public attitudes towards glass recycling in Scotland. Waste Manag. Res. 1990, 8,

177–192. [CrossRef]13. Martin, M.; Williams, I.D.; Clark, M. Social, cultural and structural influence on household waste recycling:

A case study. Resour. Conserv. Recycl. 2006, 48, 357–395. [CrossRef]14. McDonald, S.; Oates, C. Reasons for non-participation in a kerbside recycling scheme. Resour. Conserv. Recycl.

2003, 39, 369–385. [CrossRef]15. Perrin, D.; Barton, J. Issues associated with transforming household attitudes and opinions into materials

recovery: A review of two kerbside recycling schemes. Resour. Conserv. Recycl. 2001, 33, 61–74. [CrossRef]16. Tonglet, M.; Phillips, P.S.; Bates, M.P. Determining the drivers for householder pro-environmental behaviour:

Waste minimisation compared to recycling. Resour. Conserv. Recycl. 2004, 42, 27–48. [CrossRef]17. Bringhenti, J.R.; Günther, W.M.R. Social participation in selective collection program of municipal solid

waste. Eng. Sanit. Ambient. 2011, 16, 421–430. [CrossRef]18. De Feo, G.; De Gisi, S. Domestic separation and collection of municipal solid waste: Opinion and awareness

of citizens and workers. Sustainability 2010, 2, 1297–1326. [CrossRef]19. Castagna, A.; Casagranda, M.; Zeni, A.; Girelli, E.; Rada, E.C.; Ragazzi, M.; Apostol, T. 3R’s from citizens

point of view and their proposal from a case-study. UPB Sci. Bull. 2013, 75, 253–264.20. Guagnano, G.A.; Stern, P.C.; Dietz, T. Influences on attitude-behavior relationships: A natural experiment

with curbside recycling. Environ. Behav. 1995, 27, 699–718. [CrossRef]21. Noehammer, H.C.; Byer, P.H. Effect of design variables on participation in residential curbside recycling

programs. Waste Manag. Res. 1997, 15, 407–427. [CrossRef]22. Rada, E.C.; Zatelli, C.; Mattolin, P. Municipal solid waste selective collection and tourism. WIT Trans.

Ecol. Environ. 2014, 180, 187–197.23. Department of the Environment, Transport and the Regions (DETR). Less Waste More Value; Department of

the Environment, Transport and the Regions: London, UK, 1998.24. Strategy Unit. Waste Not, Want Not: A Strategy for Tacking the Waste Problem in England; Strategy Unit: London,

UK, 2002.25. Everett, J.W.; Peirce, J.J. Curbside recycling in the USA: Convenience and mandatory participation.

Waste Manag. Res. 1993, 11, 49–61. [CrossRef]

Sustainability 2016, 8, 152 14 of 14

26. Ragazzi, M.; Catellani, R.; Rada, E.C.; Torretta, V.; Salazar-Valenzuela, X. Management of municipal solidwaste in one of the Galapagos Islands. Sustainability 2014, 6, 9080–9095. [CrossRef]

27. Glenn, J. Upfront processing at MSW composting facilities. BioCycle 1991, 32, 30–33.28. Steuteville, R. The state of garbage in America. BioCycle 1995, 4, 54–63.29. Noh, M.; Lee, Y.; Oh, S.; Chu, C.; Gwack, J.; Youn, S.K.; Cho, S.H.; Lee, W.J.; Huh, S. Spatial and temporal

distribution of plasmodium vivax malaria in Korea estimated with a hierarchical generalized linear model.Osong Public Health Res. Perspect. 2012, 3, 192–198. [CrossRef] [PubMed]

30. Yau, Y. Domestic waste recycling, collective action and economic incentive: The case in Hong Kong.Waste Manag. 2010, 30, 2440–2447. [CrossRef] [PubMed]

31. Minton, A.P.; Rose, R.L. The effects of environmental concern on environmentally friendly consumerbehavior: An exploratory study. J. Bus. Res. 1997, 40, 37–48. [CrossRef]

32. Desa, A.; Kadir, N.B.A.; Yusooff, F. A study on the knowledge, attitudes, awareness status and behaviourconcerning solid waste management. Procedia Soc. Behav. Sci. 2011, 18, 643–648. [CrossRef]

33. Miafodzyeva, S.; Brandt, N. Recycling behaviour among householders: Synthesizing determinants viameta-analysis. Waste Biomass Valoriz. 2013, 4, 221–235. [CrossRef]

34. Afroz, R.; Masud, M.M.; Akhtar, R.; Duasa, J.B. Survey and analysis of public knowledge, awareness andwillingness to pay in Kuala Lumpur, Malaysia: A case study on household WEEE management. J. Clean. Prod.2013, 52, 185–193. [CrossRef]

35. Zhang, H.; Wen, Z.G. Residents’ household solid waste (HSW) source separation activity: A case study ofSuzhou, China. Sustainability 2014, 6, 6446–6466. [CrossRef]

36. Wellar, B.S. Information for Decision—Making by Rural Public Authorities; OECD: Williamsburg, VA, USA, 1984.37. Guerra, S. Markets for Recyclables: The Challenge for Local Government Recycling Programs; ICMA: Washington,

DC, USA, 1992.38. De Feo, G.; De Gisi, S. Public opinion and awareness towards MSW and separate collection programmes:

A sociological procedure for selecting areas and citizens with a low level of knowledge. Waste Manag.2010, 30, 958–976. [CrossRef] [PubMed]

39. Derksen, L.; Gartrell, J. The social context of recycling. Am. Sociol. Rev. 1993, 58, 434–442. [CrossRef]40. Timlett, R.E.; Williams, D. Public participation and recycling performance in England: A comparison of tools

for behaviour change. Resour. Conserv. Recycl. 2008, 52, 622–634. [CrossRef]41. Ando, A.W.; Gosselin, A.Y. Recycling in multifamily dwellings: Does convenience matter? Econ. Inq. 2005, 43,

426–438. [CrossRef]42. Sidique, S.F.; Lupi, F.; Joshi, S.V. The effects of behavior and attitude on drop-off recycling activities.

Resour. Conserv. Recycl. 2010, 54, 163–170. [CrossRef]43. Bernstad, A. Household food waste separation behavior and the importance of convenience. Waste Manag.

2014, 34, 1317–1323. [CrossRef] [PubMed]44. Darwiche, A. Modeling and Reasoning with Bayesian Networks, 1st ed.; Cambridge University Press: London,

UK, 2009.45. Jensen, F.V.; Nielsen, T.D. Bayesian Networks and Decision Graphs, 2nd ed.; Springer Verlag: Heidelberg, Berlin,

Germany, 2007.46. Chu, Z.J.; Xi, B.; Song, Y.; Crampton, E. Taking out the trash: Household preferences over municipal solid

waste collection in Harbin, China. Habitat Int. 2013, 40, 194–200. [CrossRef]47. Harbin Municipal Bureau of Statistics of China. Harbin Statistical Yearbook; China Statistics Press: Beijing,

China, 2014.

© 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open accessarticle distributed under the terms and conditions of the Creative Commons by Attribution(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).